Seediscussions,stats,andauthorprofilesforthispublicationat:https://www.researchgate.net/publication/258806181

InterplaybetweenCrystallizationandPhaseSeparationinPS-b-PMMA/PEOBlends:TheEffectofConfinement

ArticleinMacromolecules·May2013

ImpactFactor:5.8·DOI:10.1021/ma4005692

CITATIONS

7

READS

79

3authors,including:

Xu-MingXie

TsinghuaUniversity

179PUBLICATIONS1,844CITATIONS

SEEPROFILE

Availablefrom:Xu-MingXie

Retrievedon:27June2016

Interplay between Crystallization and Phase Separation inPS‑b‑PMMA/PEO Blends: The Effect of ConfinementNan Chen, Li-Tang Yan, and Xu-Ming Xie*

Key Laboratory of Advanced Materials (MOE), Department of Chemical Engineering, Tsinghua University, Beijing 100084, China

*S Supporting Information

ABSTRACT: Interplay between phase separation and crystallization underconfinement for the blends of PEO homopolymers with different molecularweight and PS-b-PMMA block copolymer is studied. Phase structures of theblends are investigated by atomic force microscope (AFM) and theoreticallysimulated by the dissipative particle dynamics (DPD) method, and a phasediagram describing the phase structure is established. Low molecular weightPEO (PEO2) disperses uniformly in the PMMA block domain and causes atransition from cylinder phase to perforated lamellar phase, while highmolecular weight PEO (PEO20) causes expansion of the cylinder domains andformation of disordered domains. Crystallization and melting behavior of theblends are detected by differential scanning calorimetry (DSC). The resultsshow the liquid−liquid phase separation between PEO homopolymer andPMMA block under PS-b-PMMA microphase-separated structure is suppresseddue to the hard confinement caused by glassy PS block. As a result, in the blends of PS-b-PMMA/PEO2, PEO2 is unable tocrystallize, and in the blends of PS-b-PMMA/PEO20, PEO20 shows a more obvious melting point depression compared with thehomopolymer blends of PMMA/PEO20.

■ INTRODUCTION

Confinement has very interesting effects on polymer systemswith respect to both the crystallization and phase separation ofpolymers. As the development of nanotechnology, phasetransition behavior of such a confined system has become atopic of extensive research. For example, when polymermixtures are confined in thin film, on one hand, the interplaybetween wetting and phase separation gives rise to the surface-directed spinodal decomposition (SDSD);1−3 on the otherhand, confinement can also induce miscibility in polymerblends due to entropic inhibition of phase separation intomicelles and slowing down of chain diffusions.4−6

Besides phase separation, crystallization of polymers underconfinement has also drawn much attention. Differentapproaches have been carried out to create 1D, 2D, and 3Dconfined environments, including self-assembly of blockcopolymers7−20 and homopolymers confined in nano-pores.21−23 It has been found that crystallization of polymersunder confinement exhibits lower melting points, differentcrystallization kinetics, and crystal orientation.7−23 Althoughconfined crystallization or phase separation has beenextensively studied separately, there are few studies on confinedsystems in which crystallization is coupled with phaseseparation.Poly(ethylene oxide) (PEO) and poly(methyl methacrylate)

(PMMA) can compose mixtures with phase diagram of uppercritical solution temperature (USCT).24−35 Recently, crystal-lization and phase separation of this blend system are ofparticular interest due to its dynamically asymmetric nature

caused by the large glass transition temperature (Tg) differ-ence.30−35 Various crystallization behaviors and morphologiescan be observed based on different quench depth and pathways.The crystallization of PEO is frustrated when it couples with asimultaneous viscoelastic phase separation.30,31

In this study, we try to further elucidate the confinementeffect on this system by blending PEO with a block copolymer,polystyrene-b-poly(methyl methacrylate). PS-b-PMMA is typ-ical block copolymer and can self-assemble into periodicstructures with well-defined size of nanometer scale.19,36,37 Theglassy PS blocks can act as “hard walls” to confine the mixturesof PMMA blocks and PEO homopolymers. Previousstudies15,16,38−42 on such block copolymer/homopolymer(AB/H) system mainly focus on the crystallization behavior,and most studies used homopolymers with repeating unitsidentical to one of the blocks of the block copolymer (AB/HA

system) due to the limited numbers of miscible polymer pairs.Studies on homopolymers that are different from either of theblocks but miscible with only one of the blocks (AB/HC

system) are relatively rare.16,38 According to previousstudies,19,43−46 the solubility of homopolymer in the micro-domains of block copolymer is significantly influenced by theratio of the molecular weight of homopolymer (MH) to that ofthe copolymer block (MA) miscible with the homopolymer(MH/MA): for low MH/MA values (<1), the homopolymer may

Received: March 17, 2013Revised: April 10, 2013Published: April 22, 2013

Article

pubs.acs.org/Macromolecules

© 2013 American Chemical Society 3544 dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−3553

be uniformly distributed in microdomains; as MH/MA increases(∼1), the homopolymer will be confined within the micro-domains; further increase of MH/MA (≫1) will lead tomacrophase separation. However, in the AB/HC system, thesituation is more complicated. One issue often being neglectedin the previous studies is that the phase separation betweenhomopolymer HC and A block occurs in a preformed andconfined microdomain. Crystallization behavior of homopol-ymer HC is closely related to the miscibility between HC and Ablock. Hence, for blends of one block copolymer and onecrystalline homopolymer, the final phase structure andcrystallization behavior are the result of interplay between thefollowing three processes (schematically shown in Figure 1):

(i) microphase separation of the block copolymer AB,producing a confined environment; (ii) liquid−liquid phaseseparation between the homopolymer HC and one block underconfinementthe phase separation ability can be tuned bychanging the molecular weight of HC; (iii) the crystallization ofhomopolymer under confinement. It will be very interesting toinvestigate how these processes interplay with each other. Theeffect of confinement on liquid−liquid phase separation andcrystallization can be studied by comparing the confined systemAB/HC with the unconfined system A/HC.

■ EXPERIMENTAL SECTIONPoly(ethylene oxide) samples with different molecular weight (PEO2(Mn = 2000 g/mol) and PEO20 (Mn = 20 000 g/mol)) werepurchased from Aldrich. The asymmetric poly(styrene-b-methylmethacrylate) (PS-b-PMMA) was purchased from Polymer Source,Inc., with total number-average molecular weight Mn of 77 000 (MPS =55 000 g/mol and MPMMA = 22 000 g/mol) and polydispersity indexMw/Mn = 1.09. For comparison, an ATRP-synthesized PMMA ofalmost identical molecular weight (Mn = 20 000 g/mol, PDI = 1.13)with the PMMA block in the PS-b-PMMA was used to preparePMMA/PEO blends.A predetermined amount of PEO and PS-b-PMMA were dissolved

in benzene with total polymer concentration of 5 mg/mL and stirredovernight. DSC samples were prepared by direct casting benzenesolutions onto a PTFE dish at room temperature. AFM samples wereprepared by casting 10 μL solutions on cleaned silicon wafers. The filmthickness was determined to be ca. 200 nm by a scratching method onAFM. To remove the residual solvent, the films were dried undervacuum at room temperature for 48 h. In order to compare withPMMA/PEO blends, the weight fraction of PEO in PS-b-PMMA/PEO blends is defined as

ϕ =+m

m m(PEO)

(PEO) (PMMA block)PEO

The corresponding weight fraction of PEO in the whole PS-b-PMMA/PEO blends is

ϕϕ ϕ

=+ ‐ ‐

=+ −

Fm

m m b(PEO)

(PEO) (PS PMMA)

(1 )/(22/77)

PEO

PEO

PEO PEO

The series of PS-b-PMMA/PEO blends with different compositionsare shown in Table 1.

AFM images were obtained on a SPM-9700, Shimadzu Inc., Japan,in tapping mode. A silicon microcantilever (spring constant 42 N/mand resonance frequency 320 kHz, Nanoworld AG, Switzerland) witha pyramid tip (radius of curvature ∼8 nm) was used for scan. Theheight images and phase images were obtained simultaneously. Theoperating point ratio (the ratio of operating point amplitude to thefree oscillation amplitude) was changed between 0.2 and 0.6 (hardtapping) to acquire the best contrast of phase image.

Differential scanning calorimeter (DSC) measurements wereperformed on a Shimadzu DSC-60 calibrated with standard indiumand zinc, employing a scan rate of 10 °C/min. The Xc of each sampleof was calculated by eq 1:

=Δ

ΔX

HF Hc

m

PEO m0 (1)

■ SIMULATION METHODSTo investigate the phase morphology and evolution kinetics ofthe blends composed of diblock copolymer and homopolymer,a mesoscale method, dissipative particle dynamics (DPD),47 isused to simulate the blends of diblock copolymer (A3B7)/homopolymer (Hp) with various volume fractions and chainlengths of Hp. Corresponding to the experiment, A denotesPMMA block, B denotes PS block, and Hp denotes PEOhomopolymers with different molecular weight. DPD is acontinuum simulation technique in three dimensions andcorrectly represents the hydrodynamic interactions. It has beensuccessfully applied to study the mesophase formation of blockcopolymers with various molecular architectures.48−50 As acoarse-grained approach, DPD can model physical phenomenaoccurring at larger time and spatial scales than typical moleculardynamics as it utilizes a momentum-conserving thermostat andsoft repulsive interactions between the beads representingclusters of molecules.In the present simulations, a bead i at position ri surrounded

by beads j ≠ i at rj (distance vector rij = ri − rj and unit vector eij= rij/rij with rij = |rij|) experiences a force with the componentsof conservative interaction force FC, dissipative force FD,random force FR, and bond force FS, i.e., f i = ∑j≠i(Fij

C + FijD + Fij

R

+ FijS), where the sum runs over all beads j.47,48 The

conservative force is given by FijC = αijω

C(rij)eij, where αij isthe maximum repulsion parameter between beads i and j. Therelation between αij and the Flory−Huggins χ-parameter isgiven by αij ≈ αii + 3.27χij. Here the interaction parameterbetween two same bead type αii is chosen to be 25 in thegeneral DPD simulation. Thus, according to the relativeinteraction strength between every two species included inthe experimental system, we can determine the parameters ofαij used in the simulations. αAB = 40, αBH = 43.3 and αAH = 25.7

Figure 1. Three possible processes in AB/HC blends.

Table 1. Composition of PS-b-PMMA/PEO Blends

ϕPEO 1.00 0.90 0.80 0.70 0.60 0.50 0.45 0.40 0.35 0.30FPEO 1.00 0.72 0.53 0.40 0.30 0.22 0.19 0.16 0.13 0.11

Macromolecules Article

dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−35533545

are set for the other interactions according the real Flory−Huggins parameters of χPS−PMMA, χPS−PEO, and χPEO−PMMA,respectively. The subscripts of A, B denote the beads of eachblock in the A3B7 diblock copolymer and the subscript of Hdenotes the beads of the homopolymer.50 Clearly, the A blockhas strong affinity to the homopolymer while the B block hasstrong repulsion to the homopolymer. The weight functionωC(rij) is chosen as ωC(rij) = 1 − rij/rc for rij < rc and ω

C(rij) = 0for rij ≥ rc, where rc is the truncate distance. The random forceFijR and the dissipative force Fij

D are given by FijR =

ωσ(rij)ξijΔt−1/2eij and FijD = −1/2σ2ω(rij)(vij·eij)eij, where vij =

vi − vj and vi denotes the velocity of bead i. ξij is a randomnumber which has zero mean and unit variance. The noiseamplitude, σ, which is an inherent parameter of the DPDmethod, indicates the strength of the noise in the random forceand also the strength of the dissipative force. The interaction ofthese both forces leads to the thermostat of the system. In thegeneral DPD simulation, σ is usually fixed at 3. The bondsbetween beads in the polymer chain are represented by Fij

S =Crij with a stiffness constant C = −4. Since the bead−beadinteractions in DPD are soft,51 the crystallization ofhomopolymer (PEO) will not be simulated. This may lead tosome difference between the simulation results and exper-imental observations of blends with high weight fraction ofPEO.Here we use a modified velocity-Verlet algorithm due to

Groot and Warren to solve the motion equation.47 The radiusof interaction, bead mass, and temperature are set as the unit,i.e., rc = m = kBT = 1. A characteristic time scale is then definedas τ = (mrc

2/kBT)1/2. To consider the effects of the ratio

between the chain length of the diblock copolymer and that ofthe homopolymer Hp, the chain length of Hp, p, is increasedfrom 1 to 6. The volume fraction of the homopolymer is alsochanged so that the total volume fraction of the homopolymerand the A block, ϕh, is increased from 0.3 to 0.5, correspondingto the compositions used in the experiments. A 3D box of sizeof (20rc)

3 is chosen in the simulations, which is large enough toavoid the finite size effects.49 A bead number density of 3/rc

3 isused. The time step of Δt = 0.02τ is chosen, ensuring theaccurate temperature control for the simulation systems.50

■ RESULTS AND DISCUSSION

Phase Structure of PS-b-PMMA/PEO Blends. Figure 2shows the AFM height and phase images of a neat PS-b-PMMAfilm. Since the weight fraction of PMMA in the blockcopolymer is about 30%, a cylindrical microphase separationstructure is expected. Figure 3 shows the AFM phase images ofthin films of blends of PS-b-PMMA with various weight

fractions (ϕPEO2= 0.3−0.5) of PEO2. Discontinuous and shortlamellar microdomains can be observed in the blends of ϕPEO2= 0.30. As ϕPEO2 increases to 0.35, lamellar microdomainsbecome longer and regular dark dots arranged on the extensionline of the lamella can be seen. Similar to previousobservations,52−55 we identify this pattern as a perforatedlamella phase. Increasing ϕPEO2 to 0.45 leads to the develop-ment of more regular, lamellar microdomains, as shown inFigure 3c,d. Since low molecular weight PEO2 has a goodmiscibility with the PMMA block, it solubilizes uniformly intothe PMMA domains and causes the transition of cylinder phaseto lamella phase. Similar evolution of microdomain structurehas also been observed in our previous study on the blends ofPVCH−PE−PVCH/paraffin15 systems. At ϕPEO2 = 0.50, brightregion with size of several hundred nanometers corresponds tothe PEO2 domain, indicating macrophase separation. The darkmatrix shows almost disordered microphase structure.As shown in Figure 4, PS-b-PMMA/PEO20 blends present

very different phase structure compared to PS-b-PMMA/PEO2

blends. Blends of ϕPEO20 = 0.30 show regular cylindermicrodomains in Figure 4a. Increase of ϕPEO20 to 0.35 leadsto expansion and connection of the cylinder domains in Figure4b. Based on an earlier and systematical study by Russell’sgroup,19 the expansion of cylinders indicates the localization ofhigh molecular weight homopolymers. At ϕPEO20 = 0.40,

Figure 2. AFM height (left) and phase images (right) of neat PS-b-PMMA film. The scale bar is 250 nm.

Figure 3. AFM phase images of PS-b-PMMA/PEO2 blends withvarious weight fraction of PEO2 (ϕPEO2): (a) 0.30, (b) 0.35, (c) 0.40,(d) 0.45, and (e) 0.50.

Figure 4. AFM phase images of PS-b-PMMA/PEO20 blends withvarious weight fraction of PEO20 (ϕPEO2): (a) 0.30, (b) 0.35, (c) 0.40,(d) 0.45, and (e) 0.50.

Macromolecules Article

dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−35533546

besides the cylinder domain, some disordered domains areobserved, indicating further aggregation of PEO20. Moredisordered domains are formed at ϕPEO20 = 0.45. WhenϕPEO20 reaches 0.50, macrophase separation takes place, whichis similar to the situation of PS-b-PMMA/PEO2 blends.Simulation Results for the Phase Morphology and

Kinetics of Diblock/Homopolymer Blends. Computersimulations can present the advantage of the morphologiesforming spontaneously when started from random config-urations and thereby proved a unique approach to explore themechanism for the phase formation of the systems undergoingmicroscopic and/or macroscopic phase separation. Althoughthe block copolymer/homopolymer systems have beenconsidered by various simulation techniques,56,57 there arefew studies giving a complete phase diagram for themorphologies of the diblock copolymer/homopolymer systemswith various volume fractions and chain lengths of thehomopolymer. Herein, we perform a large scale of DPDsimulations for the systems composed of A3B7 diblockcopolymer and homopolymer Hp. The volume fraction andthe chain length of the homopolymer are changed in order toexamine the influences of these two parameters on themorphologies of the systems and, furthermore, to gain insightinto the mechanism of the morphology transition directed bythe homopolymer. Although there is no glass transition or theviscoelastic asymmetry in the DPD simulation, it is verypowerful in determining the phase morphologies, which havebeen confirmed in various simulation studies for variouspolymer systems. Thereby, through the application of DPDsimulations, we can determine the phase morphology in a morefull view, which greatly facilitates our understanding for theexperimental results of the phase morphologies and thecrystalline behaviors. Moreover, a phase diagram indicatingthe relationships of the morphologies and these two parametersis drawn based on the simulation results. It should be pointedout that this phase diagram is obtained based on the dynamicalprocess of the systems instead of the thermodynamicallyequilibrium calculation. As some structures predicted by thethermodynamic theory cannot or are practically difficult torealize due to the presence of metastable state induced by theenergy barrier, the phase diagram based on the dynamicalprocess will facilitate the comparison with the experimentalobservations. In particular, some new morphologies, e.g., thethick cylinders and the coexistence of disordered phase andcylinders, are identified in the simulations, which will bedemonstrated below. Furthermore, the evolution kinetics ofsystems is also considered in the study, which should providesome useful information for the understanding of theexperimental observations.Figure 5 shows the typical simulated structures of the A3B7/

Hp polymer blends with the chain length of the homopolymer,p, increasing from 1 to 3. According to Groot et al.’s study,48

hexagonal cylinders form via a metastable gyroid-like structureand therefore correspond to the nucleation-and-growthmechanism, whereas the lamellar phase is formed via a spinodaldecomposition. Although the hexagonal cylindrical structure isdefinitely demonstrated in Figure 5a, there are still some thintubes connecting neighboring cylinders. The appearance of theconnected tubes indicates that the system gets stuck into a localminimum in free energy surface.49 A prohibited long time maybe needed for the system to jump over this energy barrier.Figure 5b−e shows the structures of the systems at p = 1,

where the total volume fraction of the homopolymer and the A

block, ϕh, is increased from 0.3 to 0.5 by changing the fractionof the homopolymer. One can find that the cylindrical structureturns to the perforated lamellar structures from ϕh = 0.3 to0.45. At ϕh = 0.5, regular lamellae form in the system. Clearly,the structures with all these compositions are ordered at p = 1.The structures of the systems with the same compositions at p= 2 are displayed in Figure 5f−i. In this case, the systems stillexhibit perforated lamellae at ϕh = 0.3 and 0.35 (see Figure5f,g). However, the microdomains seem to be more separatedcompared to the same composition at p = 1 (see Figure 5b,c).At ϕh = 0.45, a coexistence of the disordered phase and thecylinders (as marked by the red circle) forms in the system,implying that the ordered structures tends to be destroyed withthe further increase of ϕh (see Figure 5h). Instead of theformation of the regular lamellae, the structure is absolutelydisordered at ϕh = 0.5 as shown in Figure 5i. Figure 5j−mpresents the structures of the systems with the samecompositions at p = 3. At ϕh = 0.3, the microdomains keepseparating so that the perforated lamellae evolves intohexagonal cylinders (see Figure 5j). However, in contrast tothe structures at p = 2, the structures at p = 3 exhibit moredisordered from ϕh = 0.35. In this case, the appearance of thecoexistence of disordered phase and cylinders is advanced to ϕh= 0.35 (see Figure 5k). It is interesting that a specialmesostructure, characterized by thick, separated, and disor-dered cylinders, is observed at ϕh = 0.45 and p = 3 (see Figure5l). A similar structure is also observed in other compositions atdifferent homopolymer chain lengths, which can be found inthe below phase diagram (Figure 6) of the phase morphology.This structure is different with the absolutely disorderedstructure as shown in Figure 5m (ϕh = 0.5 and p = 3) becausethe thick cylinders can still be observed although theirdiameters present large diversity. Actually, the experimentobservation demonstrates that the surface roughness (Ra) of ϕh= 0.45 is much smaller than that of ϕh = 0.5.

Figure 5. Typical simulated mesostructures in the bulk of the diblockcopolymer A3B7 without (a) and with homopolymer Hp (b−m). Thelength of the homopolymer chain, p, of each snapshot is p = 1 (b−e), p= 2 (f−i), and p = 3 (j−m). The total volume fractions of block A andHp are ϕh = 0.3 (b, f, j), ϕh = 0.35 (c, g, k), ϕh = 0.45 (d, h, l), and ϕh =0.5 (e, i, m). The isosurface of the block copolymer at an orderparameter of zero is colored yellow, indicating the interface betweenblock B and other components. The green regions mark phase A andHp, and the transparent regions indicate phase B. The homopolymers,Hp, are represented by the purple beads. The red circles mark thecylindrical structures in the corresponding snapshots.

Macromolecules Article

dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−35533547

Figure 6 is a phase diagram showing the simulated meso-structures of the A3B7/Hp polymer blends in response tovarious volume fractions and chain lengths of the homopol-ymer. The phase diagram is obtained based on the dynamicalprocess of the phase evolution instead of the thermodynami-cally equilibrium theory. Thereby the potential metastablestates formed due to the local energy barrier may be included init, which, however, leads to a better accordance with theexperiment observation compared to the phase diagramobtained by the thermodynamically equilibrium calculation.From this phase diagram, one can find that the orderedstructures occur in the systems with low volume fraction andshort chain length of the homopolymer. Increasing either the

volume fraction or the chain length of the homopolymerinduces the morphology transition of the blends. Weconsequently find the formation of ordered structures includingperforated lamellae, lamellae, hexagonal cylinders, andcylinders. When the volume fraction and the chain length ofthe homopolymer are increased to a certain extent, the orderedstructures are destroyed. In this case, the phase sections for thecoexistence of disordered phase and cylinders, thick cylinders,and disordered phase are present in the phase diagram. By thistoken, the phase diagram gives a complete view on themorphology transition of the diblock copolymer induced by thehomopolymer.Relatively, the two compositions used in the experiments of

the study correspond to the A3B7 block copolymer with thehomopolymers of H1 (short polymer chain) and H3 (longpolymer chain or the polymer chain is comparable with thelength of the A block), respectively. It also should be pointedout that AFM provides surface information on phasemorphology and the DPD simulations give bulk data. Thebulk data of the DPD simulations provide a more detailed andclear view on the phase morphology of the blend system, whichcan help us to understand the experimental results. Basically,the simulated morphologies have a good agreement with thoseobserved in the experiments (see Figures 3 and 4). However, aspecial case takes place at ϕh = 0.5 and p = 1, where a regularlamellar structure is found in the simulation (see Figure 5e),whereas disordered structure is observed in the experiment (seeFigure 3e). In this case, the relatively high weight fraction ofPEO2 has a strong tendency to crystallize. As mentionedbefore, DPD methods only predict the phase structure ofblends in the amorphous state. In a real experiment,crystallization of PEO2 at high ϕPEO2 value destroys the regularlamellar structure, leading to macrophase separation.

Figure 6. A phase diagram showing the simulated mesostructures ofthe A3B7/Hp blends with various volume fractions ϕh and chain lengthsp of the homopolymer Hp. The morphologies are (□) perforatedlamellae, (■) lamellae, (●) hexagonal cylinders, (○) cylinders, (◎)disordered phase and cylinders, (△) thick cylinders, and (◇)disordered phase. The phase boundary redlines are drawn to guidethe eye.

Figure 7. Spatial distribution of homopolymer Hp within the A phase of A3B7 diblock copolymers at ϕh = 0.5. (a) p = 1 and (b) p = 3. These twoimages correspond to Figures 5e and 5m, respectively. The isosurface of the block copolymer at an order parameter of zero is colored yellow,indicating the interface between block B and other components. The homopolymers, Hp, are represented by the purple beads. DPD beads of A blockare not shown here. (c) and (d) show the corresponding statistics of the distance between each Hp bead (or A bead) and the phase interface of blockcopolymers; the lines are drawn to guide the eye.

Macromolecules Article

dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−35533548

To delineate the formation of the various morphologies, wealso provide some videos in the Supporting Information,showing the evolution kinetics of the blends with differentcompositions. The supporting videos 1−4 display the processesof the structure evolution and formation at ϕh = 0.3 and p = 1,ϕh = 0.4 and p = 1, ϕh = 0.3 and p = 3, and ϕh = 0.5 and p = 3,respectively. From these videos, we can find that thehomopolymer beads gradually move to their preferentialphase, although they are randomly dispersed at the initialstage. For the ordered structure, the homopolymer beadspresent a uniform distribution. However, when the volumefraction and the chain length of the homopolymer are increasedto a certain extent, these homopolymer beads will separate fromtheir preferential phase and aggregate at the center of themicrodomains, leading to the formation of the macrophaseseparation. In this case, the defects of the phase are difficult toeliminate, and the ordered structures are destroyed. Obviously,these videos give a clear and detailed description of how thehomopolymer beads are included in their preferential domains,how the structure defects of the microdomains are eliminatedwith the structure evolution, and, finally, how the ordered orthe disordered structures form at the final stage.To gain insight into the mechanism about the morphology

transition of the blends composed of diblock copolymer andhomopolymer, the spatial distribution of the homopolymerwithin their preferential microdomains is examined in the study.As an example, Figure 7a,b shows the spatial distribution of thehomopolymer beads within phase A, corresponding to theimages of Figure 5e,m. As demonstrated in Figures 5e and 7a,the homopolymers with a single bead (H1) can be uniformlydispersed in their preferential domain. However, at the sameconcentration of homopolymers, the homopolymers withlonger chains (H3) tend to aggregate and lead to macrophaseseparation (see Figures 5m and 7b). It should be noted that thetotal number of the homopolymer beads and the interactionbetween the homopolymer and the block copolymers are notchanged in these two systems. Furthermore, in DPD, the size ofthe bead representing a fluid unit is also fixed. The onlydifference is the chain length of the homopolymer, i.e., p = 1 inFigure 7a and p = 3 in Figure 7b. Thereby, it can be safelyconcluded that the morphology difference of the two systems isdue to the entropy effect because the entropy of the beadslinked into a chain (i.e., H3) is different in both shape and sizecompared to the single bead (i.e., H1). We have also quantifiedthe distribution of homopolymer beads within phase domainsfor these both figures by calculating the distance between eachhomopolymer bead (or A block bead) and the phase interfaceof block copolymers. As shown in Figure 7c,d, the distances ofall homopolymer beads and A beads to the interface are

summarized together, and their distribution in the phasedomain can be obtained based on the statistic of the data.Clearly, H1 almost uniformly distrubute in the A domain whileH3 present a much wider distribution due to their aggregationin the center of A domains with various sizes. The spatialdistribution of the inclusions in microdomains of blockcopolymers is governed by complex interplay betweenenthalpic and entropic interactions. It should be noted thatthe homopolymers have a preferential interaction (attraction)with one phase of the block copolymers. This enthalpicinteraction tends to constrain the homopolymers in thepreferential phase. For the homopolymers with short or longchains, the strengths of enthalpic interaction with blockcopolymers are not changed. The only change lies at theentropic effect, i.e., the shape and the size of thehomopolymers. Thereby, compared to the system with shorthomopolymer chains, the macrophase separation of the systemwith long homopolymer chain is induced by the change of theentropic effect. A very initial simulation by Thompson et al.58

on the basis of mesoscale simulations revealed that largerinclusions localize at the center of their preferential micro-domains, whereas smaller nanoparticles are more uniformlydispersed within a specific microdomains. O’Shaughnessy etal.59 obtained similar conclusions, using analytical scaling theoryto investigate the behavior of nanometer-scale inclusions ingrafted layers. If the inclusion size is small compared with thecontour length of their preferential block, the inclusions aredispersed within this phase, and the overall structures of thenanocomposites are dictated by the block copolymers.However, if the inclusion size is larger compared with thecontour length, the polymer must stretch around the inclusionsto accommodate them, which will lead to the penalty of theconformation entropy because the extended chains can accessfewer conformations than the relaxed coil. When the inclusionsize reaches a critical value, this cost in free energy becomes toogreat, and the inclusions can no longer disperse in theirpreferential domains; instead, the inclusions are segregated intothe center to reduce the stretching of the polymer chains.58

Although this conclusion was drawn from the block copolymerswith nanoparticles, it is still suitable for the understanding ofthe morphology transition induced by the change of thehomopolymer chain length because the radius of gyration of ahomopolymer chain is similar to the diameter of a nanoparticle.

Melting Behavior of PS-b-PMMA/PEO Blends. Themelting and glass transition behavior of PS-b-PMMA/PEO2blends is shown in Figure 8 and Table 2. The first heatingcurves on DSC reflect the phase separation state of the blendsafter the solvent evaporation. It should be noted that the phase-separated structure is not in equilibrium, especially for blends

Figure 8. DSC heating curves of PS-b-PMMA/PEO2 blends: (a) first heating, (b) after annealing at 180 °C for 1 h, (c) after annealing at 180 °C for48 h.

Macromolecules Article

dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−35533549

with high ϕPEO2. Blends of ϕPEO2 < 0.40 show no melting peakon the first heating curves in Figure 8a, indicating PEO2 tookthe amorphous state and dissolved in the PMMA micro-domains, which is consistent with the AFM observations(Figure 3a,b) and DPD simulations (Figures 5 and 6). Blendsof 0.5 > ϕPEO2 ≥ 0.40 show similar melting points and very lowcrystallinity (Table 2), indicating most PEO2 are still dissolvedin the PMMA microdomain and a small fraction of PEO2aggregate outside the PMMA domain. This is also supported bythe AFM observations of lamellar microphase structure (Figure3c,d). The blend of ϕPEO2 = 0.50 has a strong tendency tocrystallize due to macrophase separation (Figure 3e). Afterannealing at 180 °C (well above the UCST of PMMA/PEO)for 1 h, PEO2 outside the PMMA lamellar domains graduallymigrates into the inside. As a result, the dilute effect becomespronounced and the melting point depression is about 10 °C(Figure 8b). It should be noticed that the migration andredispersion of PEO2 into the PMMA domains need a certaintime to reach equilibrium. This evolution process of phasestructure with time can also be observed in the DPD simulation(see Supporting Information). The blends of 1 h annealingtime are in an intermediate state. It is also worth to note that inFigure 8c Neat PS-b-PMMA shows two Tg on DSC, 103.5 °Cfor PS block and 120.7 °C for PMMA block, while the blendsafter 48 h annealing show much lower Tg at 80−85 °C. Thedepression and broadening of Tg indicate uniform solubilizationof PEO2 in the PMMA domain. During the cooling processfrom 180 to −80 °C, the glass transition of the PS block occursfirst, and local segregation of PEO2 inside the PMMA domainis prohibited due to the confinement from the glassy PS block.The absence of the PEO-rich phase frustrates the crystallizationof PEO2, so the blends of 48 h annealing time shows nomelting peak. The confinement effect of PS block can befurther illustrated by a comparison experiment of PMMA/PEO2 blends. As shown in Figure 9, under the same annealingand crystallization conditions, homopolymer blends of PMMA/PEO2, which are free of confinement, can still crystallize.Increasing the molecular weight of PEO can usually lead to a

stronger tendency of liquid−liquid phase separation betweenPEO and PMMA (or PMMA block). In order to further clarifythe effect of confinement on the liquid−liquid phase separationof PMMA/PEO, PEO with higher MW (PEO20) was blendedwith PS-b-PMMA and homo-PMMA, respectively. TakingϕPEO20 = 0.4 as an example, Figure 10 shows similar depressionof crystallinity and melting point. It should also be noted herethat the movement of block copolymer is frozen as the solventevaporates. Thereby, annealing treatment is necessary to get the

equilibrium phase structure. However, it takes much more timefor PEO20 to completely dissolve in the PMMA domaincompared to PEO2 (Tm peak of ϕPEO2 = 0.4 disappears afterannealing for only 1 h). The double endotherm peak of PS-b-PMMA/PEO20 indicates PEO crystals under two situations.One is crystallized inside the cylinder which is diluted by morefractions of PMMA, resulting in a lower Tm according toNishi−Wang’s equation.60,61 The other is crystallized outsidethe cylinder, which exhibits a higher Tm. Annealing leads tomore PEO molecules migrating into the PMMA cylindricaldomains, which results in the decrease of melting point andcrystallinity with annealing time.Figure 11 shows the DSC curves of one annealed sample

(ϕPEO20 = 0.4, annealing time = 4 h) heating from differentstarting temperature. The upper limit of the heating programwas set at 90 °C, which is below the Tg of PS block; therefore,the microphase separation structure will be retained during theheating−cooling circles. The blends show double endothermpeak on the first heating curve: one at 53.4 °C and the other at57.9 °C. When heated from 20 °C (second heating), the lowerTm disappears while the higher one is retained. When heatingfrom −80 °C again (third heating), the lower Tm occurs again.These results indicate the supercooling needed for thecrystallization of PEO inside or outside the cylinder domainsare different. The PEO confined in the cylinders needs largersupercooling to induce homogeneous nucleation.10,23 Theresults can also be reproduced even during more circles ofheating and cooling, with Tm and heat of fusion almostunchanged, as long as the heating upper limit is not beyond 90°C. This indicates the migration of PEO molecules fromdisorder domains outside the cylinder into the cylinder isobstructed by the glassy “PS walls”. However, once the blendsare heated above the Tg of PS block, PEO can migrate into thecylinder. For example, as the annealing time at 180 °C increasesto 6 h (Figure 10a), the higher Tm almost disappears while thelower Tm still exists. As mentioned above, PEO20 has astronger tendency to phase-separate from PMMA compared toPEO2. According to the AFM and DPD results (Figures 4 and5), PEO20 tends to aggregate in the cylindrical domains, whichprovides a PEO-rich domain for PEO20 to crystallize. Furtherincreasing the annealing time to 48 h, similar to the situation ofPEO2, the Tm peaks completely disappear (Figure 10a).However, if the samples were submitted to isothermalcrystallization for a sufficient time (48 h), PEO20 can crystallize

Table 2. Melting Temperature, Crystallinity, and Tg of PS-b-PMMA/PEO2 Blends

ϕPEO2

0.50 0.45 0.40 0.35 0.30

Tm/°C (0 h) 51.8 50.3 51.9Tm/°C (1 h) 43.9/50.6 45.7Tm/°C (48 h)Xc (0 h) 0.07 0.05 0.02Xc (1 h) 0.09 0.03Xc (48 h)Tga/°C (0 h) 96.4 94.2 96.4 97.4 96.9

Tga/°C (1 h) 93.8 95.6 95.4 98.0 96.1

Tga/°C (48 h) 85.2 84.7 80.4 82.1 84.5

aTg data are determined as the onset temperature of glass transition.

Figure 9. DSC heating curves of PEO2 blended with PS-b-PMMA(dotted lines) and homo-PMMA (solid lines).

Macromolecules Article

dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−35533550

again, while PEO2 still cannot crystallize. The results are shownin Figure 12.PS-b-PMMA/PEO20 blends with various ϕPEO20 (0.3−0.9)

were then submitted to isothermal crystallization at 30 °C for48 h, and the corresponding Tm are shown in Figure 12 (hollowcircles). For comparison, the Tm of homopolymer blends(PMMA/PEO20) are also presented (solid circles). At ϕPEO20

≤ 0.30, both PS-b-PMMA/PEO20 and PMMA/PEO20 showno traces of crystallization; the phase morphology of PS-b-PMMA/PEO20 is dominated by cylindrical structure (Figure12, inset a). At ϕPEO20 > 0.5, these two series of blends showsimilar melting point depression. In this range, macrophaseseparation in PS-b-PMMA/PEO20 takes place, and thecrystallization of PEO20 breaks through the confinement;take ϕPEO20 = 0.6 as an example, typical dendritic crystals ofPEO20 can be observed on AFM (Figure 12, inset b). At 0.30 <ϕPEO20 < 0.5, the Tm of PMMA/PEO20 are almost constant(ca. 62 °C); however, the Tm of PS-b-PMMA/PEO20 blendscontinues to decrease with the decrease of ϕPEO20.

According to the study by Li et al.25 in 1984 and recentstudies by Han’s group30−33 on PMMA/PEO blends,crystallization of PEO is accompanied by the liquid−liquidphase separation between PEO and PMMA. A PEO-rich phasegenerated from liquid−liquid phase separation is required forcrystallization, and crystallization of PEO can also influence thephase separation in reverse. According to the phase diagramgiven by Shi et al.,33 in the region of low PEO weight fraction,crystallization kinetics is greatly depressed; thus, the liquid−liquid phase separation may occur prior to crystallization. Thisexplains the disappearance of melting point depression ofPMMA/PEO20 at 0.30 < ϕPEO20 < 0.50. As for the PS-b-PMMA/PEO20 blends, however, the liquid−liquid phaseseparation of PEO and PMMA block is confined in thepreformed microdomains. The movement of PMMA blocks isconfined by glassy PS blocks; therefore, the liquid−liquidseparation generates PEO-rich phases containing a considerableamount of PMMA blocks. In other words, the dilute effect ofPMMA on the melting of PEO crystal under confinement is

Figure 10. (a) DSC second heating curves of PS-b-PMMA/PEO20 blends (ϕPEO20 = 0.4) with different annealing time. (b) Corresponding heatingprogram.

Figure 11. (a) DSC heating curves of PS-b-PMMA/PEO20 blends (ϕPEO20 = 0.4, annealing time = 4 h) from different starting temperature. (b)Corresponding heating program.

Macromolecules Article

dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−35533551

intensified. This confined phase separation and crystallizationprocess is sketched in Figure 13.

■ CONCLUSIONThis work studied the interplay between crystallization andphase separation in the blends of PS-b-PMMA/PEO. At ϕPEO2<0.5, low molecular weight PEO (PEO2) dissolves uniformly inPMMA microdomains, causing a transition of microphaseseparation structure from cylinders to perforated lamellas andfurther to lamellas with increasing PEO weight fraction, whilehigh molecular weight PEO (PEO20) aggregates at the centerof cylindrical domains, causing an expansion of cylinder andfurther formation of disorder domains with increasing PEOweight fraction. DPD simulation provides a phase diagram thathelps better understanding the influences of molecular weightand weight fraction of homopolymer on the phase structure ofblock copolymer/homopolymer blends.In the blends of PS-b-PMMA/PEO, the liquid−liquid phase

separation between PEO and PMMA block is confined by theglassy PS block. As a result, low MW PEO (PEO2) is unable tocrystallize due to the complete suppression of liquid−liquidphase separation; high MW PEO (PEO20) has a strongerphase-separation ability and can still crystallize. Because of

confinement, PS-b-PMMA/PEO20 shows more obviousmelting point depression compared with PMMA/PEO20.

■ ASSOCIATED CONTENT

*S Supporting InformationVideos showing the processes of the structure evolution andformation simulated by the DPD method. This material isavailable free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATION

Corresponding Author*E-mail: xxm-dce@mail.tsinghua.edu.cn.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work is supported by National Natural ScienceFoundation of China (No. 20874056).

■ REFERENCES(1) Binder, K.; Puri, S.; Das, S. K.; Horbach, J. J. Stat. Phys. 2010, 138,51−84.(2) Yan, L. T.; Li, J. L.; Li, Y.; Xie, X. M. Macromolecules 2008, 41,3605−3612.(3) Yan, L. T.; Xie, X. M. J. Chem. Phys. 2007, 126, 064908.(4) Hsu, W. P. Colloids Surf., A 2010, 367, 161−166.(5) Jiang, K.; Su, Y.; Xie, B.; Meng, Y.; Wang, D. J. Phys. Chem. B2009, 113, 3269−3272.(6) Zhu, S.; Liu, Y.; Rafailovich, M. H.; Sokolov, J.; Gersappe, D.;Winesett, D. A.; Ade, H. Nature 1999, 400, 49−51.(7) Cheng, S. Z. D.; Hsiao, M. S.; Zheng, J. X.; Horn, R. M. V.;Quirk, R. P.; Thomas, E. L.; Chen, H. L.; Lotz, B. Macromolecules2009, 42, 8343−8352.(8) Hsiao, M. S.; Chen, W. Y.; Zheng, J. X.; Van Horn, R. M.; Quirk,R. P.; Ivanov, D. A.; Thomas, E. L.; Lotz, B.; Cheng, S. Z. D.Macromolecules 2008, 41, 4794−4801.(9) Nojima, S.; Ohguma, Y.; Kadena, K.-i.; Ishizone, T.; Iwasaki, Y.;Yamaguchi, K. Macromolecules 2010, 43, 3916−3923.(10) Nojima, S.; Ohguma, Y.; Namiki, S.; Ishizone, T.; Yamaguchi, K.Macromolecules 2008, 41, 1915−1918.(11) Sun, Y. S.; Chung, T. M.; Li, Y. J.; Ho, R. M.; Ko, B. T.; Jeng, U.S.; Lotz, B. Macromolecules 2006, 39, 5782−5788.(12) Yu, P. Q.; Xie, X. M.; Wang, T.; Bates, F. S. J. Appl. Polym. Sci.2006, 102, 2584−2589.(13) Yu, P. Q.; Xie, X. M.; Wang, Z.; Li, H. S.; Bates, F. S. Polymer2006, 47, 1460−1464.(14) Chen, Y. F.; Zhang, F. B.; Xie, X. M.; Yuan, J. Y. Polymer 2007,48, 2755−2761.(15) Yu, P. Q.; Yan, L. T.; Chen, N.; Xie, X. M. Polymer 2012, 53,4727−4736.(16) Ho, R. M.; Chiang, Y. W.; Lin, C. C.; Huang, B. H.Macromolecules 2005, 38, 4769−4779.(17) Park, S. C.; Jung, H.; Fukukawa, K. I.; Campos, L. M.; Lee, K.;Shin, K.; Hawker, C. J.; Ha, J. S.; Bang, J. J. Polym. Sci., Polym. Chem.2008, 46, 8041−8048.(18) Torikai, N.; Takabayashi, N.; Noda, I.; Koizumi, S.; Morii, Y.;Matsushita, Y. Macromolecules 1997, 30, 5698−5703.(19) Jeong, U.; Ryu, D. Y.; Kho, D. H.; Lee, D. H.; Kim, J. K.;Russell, T. P. Macromolecules 2003, 36, 3626−3634.(20) Mayes, A. M.; Russell, T. P.; Satija, S. K.; Majkrzak, C. F.Macromolecules 1992, 25, 6523−6531.(21) Shin, K.; Woo, E.; Jeong, Y. G.; Kim, C.; Huh, J.; Kim, K. W.Macromolecules 2007, 40, 6617−6623.(22) Woo, E.; Huh, J.; Jeong, Y. G.; Shin, K. Phys. Rev. Lett. 2007, 98,136103.

Figure 12. Comparison of melting point depression in PS-b-PMMA/PEO20 blends and PMMA/PEO20 blends.

Figure 13. Schema of confined phase separation and crystallization inPS-b-PMMA/PEO blends.

Macromolecules Article

dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−35533552

(23) Duran, H.; Steinhart, M.; Butt, H. J.; Floudas, G. Nano Lett2011, 11, 1671−1675.(24) Ferreiro, V.; Douglas, J. F.; Amis, E. J.; Karim, A. Macromol.Symp. 2001, 167, 73−88.(25) Li, X.; Hsu, S. L. J. Polym. Sci., Polym. Phys. 1984, 22, 1331−1342.(26) Martuscelli, E.; Silvestre, C.; Addonizio, M. L.; Amelino, L.Makromol. Chem. 1986, 187, 1557−1571.(27) Mu, D.; Huang, X. R.; Lu, Z. Y.; Sun, C. C. Chem. Phys. 2008,348, 122−129.(28) Okerberg, B. C.; Marand, H. J. Mater. Sci. 2007, 42, 4521−4529.(29) Suvorova, A. I.; Hassanova, A. H.; Tujkova, I. S. Polym. Int.2000, 49, 1014−1016.(30) Shi, W. C.; Han, C. C. Macromolecules 2012, 45, 336−346.(31) Shi, W. C.; Xie, X. M.; Han, C. C. Macromolecules 2012, 45,8336−8346.(32) Shi, W. C.; Yang, J.; Zhang, Y.; Luo, J.; Liang, Y. R.; Han, C. C.Macromolecules 2012, 45, 941−950.(33) Shi, W. C.; Cheng, H.; Chen, F. H.; Liang, Y. R.; Xie, X. M.;Han, C. C. Macromol. Rapid Commun. 2011, 32, 1886−1890.(34) Schwahn, D.; Pipich, V.; Richter, D. Macromolecules 2012, 45,2035−2049.(35) Brodeck, M.; Alvarez, F.; Colmenero, J.; Richter, D. Macro-molecules 2012, 45, 536−542.(36) Ham, S.; Shin, C.; Kim, E.; Ryu, D. Y.; Jeong, U.; Russell, T. P.;Hawker, C. J. Macromolecules 2008, 41, 6431−6437.(37) Bang, J.; Jeong, U.; Ryu, D. Y.; Russell, T. P.; Hawker, C. J. Adv.Mater. 2009, 21, 4769−4792.(38) Lee, J. K.; Kim, J. S.; Lim, H. J.; Lee, K. H.; Jo, S. M.; Ougizawa,T. Polymer 2006, 47, 5420−5428.(39) Huang, P.; Guo, Y.; Quirk, R. P.; Ruan, J. J.; Lotz, B.; Thomas,E. L.; Hsiao, B. S.; Avila-Orta, C. A.; Sics, I.; Cheng, S. Z. D. Polymer2006, 47, 5457−5466.(40) Zhu, L.; Mimnaugh, B. R.; Ge, Q.; Quirk, R. P.; Cheng, S. Z. D.;Thomas, E. L.; Lotz, B.; Hsiao, B. S.; Yeh, F. J.; Liu, L. Z. Polymer2001, 42, 9121−9131.(41) Huang, P.; Zhu, L.; Guo, Y.; Ge, Q.; Jing, A. J.; Chen, W. Y.;Quirk, R. P.; Cheng, S. Z. D.; Thomas, E. L.; Lotz, B.; Hsiao, B. S.;Avila-Orta, C. A.; Sics, I. Macromolecules 2004, 37, 3689−3698.(42) Floudas, G.; Tsitsilianis, C. Macromolecules 1997, 30, 4381−4390.(43) Hashimoto, T.; Tanaka, H.; Hasegawa, H. Macromolecules 1990,23, 4378−4386.(44) Tanaka, H.; Hasegawa, H.; Hashimoto, T. Macromolecules 1991,24, 240−251.(45) Winey, K. I.; Thomas, E. L.; Fetters, L. J. Macromolecules 1991,24, 6182−6188.(46) Hashimoto, T.; Izumitani, T.; Oono, K. Macromol. Symp. 1995,98, 925−949.(47) Groot, R. D.; Warren, P. B. J. Chem. Phys. 1997, 107, 4423−4435.(48) Groot, R. D.; Madden, T. J. J. Chem. Phys. 1998, 108, 8713−8724.(49) Qian, H. J.; Lu, Z. Y.; Chen, L. J.; Li, Z. S.; Sun, C. C.Macromolecules 2005, 38, 1395−1401.(50) Yan, L. T.; Popp, N.; Ghosh, S. K.; Boeker, A. ACS Nano 2010,4, 913−920.(51) Shillcock, J. C.; Lipowsky, R. J. Chem. Phys. 2002, 117, 5048−5061.(52) Horvat, A.; Lyakhova, K. S.; Sevink, G. J. A.; Zvelindovsky, A.V.; Magerle, R. J. Chem. Phys. 2004, 120, 1117−1126.(53) Knoll, A.; Magerle, R.; Krausch, G. J. Chem. Phys. 2004, 120,1105−1116.(54) Sriprom, W.; James, M.; Perrier, S.; Neto, C. Macromolecules2009, 42, 3138−3146.(55) Tsarkova, L.; Knoll, A.; Krausch, G.; Magerle, R. Macromolecules2006, 39, 3608−3615.(56) Huang, Y. M.; Xia, H.; Liu, H. L.; Hu, Y. Macromol. TheorySimul. 2007, 16, 93−100.

(57) Martinez-Veracoechea, F. J.; Escobedo, F. A. Macromolecules2009, 42, 1775−1784.(58) Thompson, R. B.; Ginzburg, V. V.; Matsen, M. W.; Balazs, A. C.Science 2001, 292, 2469−2472.(59) Kim, J. U.; O’Shaughnessy, B. Macromolecules 2006, 39, 413−425.(60) Nishi, T.; Wang, T. T. Macromolecules 1975, 8, 909−915.(61) Nishi, T.; Wang, T. T.; Kwei, T. K. Macromolecules 1975, 8,227−234.

Macromolecules Article

dx.doi.org/10.1021/ma4005692 | Macromolecules 2013, 46, 3544−35533553